Ultrasound Image Denoising Based on Adaptive Lifting Wavelet Transform

摘要

The performance of ultrasound image is generally affected by multiplicative speckle noise. In the paper, an adaptive image denoising strategy for speckle suppression within the framework of adaptive lifting wavelet transform and Bayesian estimation is proposed. Adaptive predictor design based on Bernstein filter and least mean square (LMS) algorithm is presented. The subband decomposition of logarithmically transformed ultrasound image is carried out with adaptive lifting wavelet transform. A Bayesian estimation procedure is used to restore the wavelet coefficients of the desired image. Numerical experiments show that compared with the standard filtering techniques, the proposed denoising method is more effective in terms of speckle reduction and signal preservation.